How Snorkl AI Made Daily Speaking Possible in My NYC World Language Classroom

When I walked into my first year teaching world languages in NYC Public Schools, I had the same vision most WL teachers quietly held onto: my students would speak the target language every single day. Not just occasionally. Not just when the lesson plan allowed it. Every day, non-stop. 

I pictured lively interpersonal exchanges, confident presentational speaking, and students slowly building fluency through consistent practice.

And then the NYCPS bell schedule introduced itself.

🕒 The NYCPS Schedule: 45 Minutes That Are Never Actually 45 Minutes

If you teach in NYC, you already know the truth: the schedule says 45 minutes, but the universe says otherwise.

By the time students transition, settle, grab materials, and we troubleshoot whatever tech decides to malfunction that day, I’m usually left with 38–40 real minutes to teach.

In that tiny window, I’m supposed to:

• deliver meaningful input
• model language
• run activities
• check for understanding
• differentiate
• manage behavior
• and somehow collect speaking samples

As a first‑year teacher with three preps every single day across two languages, I had to face a hard truth: Daily speaking is essential, and daily grading is impossible.

The Problem With “Just Collect Speaking Samples”

Because time is tight, student recordings often end up being 10–20 seconds long. But shorter recordings don’t make feedback easier — they make it harder.

Every second matters. Every word carries weight. Every attempt at meaning deserves attention.

To give feedback that is:

• meaningful
• accurate
• proficiency‑aligned
• supportive
• and actionable

I would need hours I simply don’t have.

If I tried to personally listen to and grade every speaking sample from every student, across three different courses, every day, I’d be buried under a mountain of audio files by October.

Enter Snorkl AI: A Daily Digital Speaking Coach

Snorkl didn’t replace me. It extended me.

It gave my students something I could never sustainably provide on my own: a patient, consistent, always‑available speaking partner that listens and responds instantly.

Here’s what daily speaking looks like now:

1. Students complete a short speaking prompt in Snork from Chromebook. 
2. Snorkl AI listens and evaluates only their speech — not background noise, not peers.
3. Students get immediate glows and growth steps written in warm, student‑friendly language. And they can reattempt with actionable feedback. 
4. I check the dashboard to see patterns and who needs support.
5. We move on — no grading backlog, no guilt, no audio files haunting me.

🛠️ Aligned with ACTFL - Use your OWN rubric

Snorkl’s built‑in rubric generator is fine as a starting point, but it’s way too vague for real proficiency‑based grading. It doesn’t clearly define what a 1, 2, 3, or 4 actually means, and it definitely doesn’t map onto ACTFL performance descriptors.

So I built my own.

I actually trained Copilot to help me design a rubric that matches exactly how I assess Novice‑level speaking. I told it:

• 3 = target proficiency
• 4 = strong performance with features of the next level
• 2 = developing Novice‑Mid
• 1 = beginning performance needing support

Copilot helped me refine the language until it was crystal clear, student‑friendly, and ACTFL‑aligned. Now Snorkl uses my rubric — not a generic one — to score student speaking.

This matters because:

• A 3 is full credit in my gradebook.
• A 4 is a bonus for students showing glimpses of Intermediate‑Low.
• A 2 tells me the student is on the right track but needs clarity.
• A 1 tells me the student needs intervention or scaffolding.

And because the rubric is so clear, Snorkl’s feedback is consistent with my values: communication over perfection, clarity over complexity, and growth over grammar policing.

Here's a sample Snorkl Rubric for interpersonal speaking:

📊 How I Use Snorkl’s 1–4 Scale in My Gradebook

Here’s how the scores translate in my classroom:

• 4 — Strong
  • Clear, confident, comprehensible speaking.
  • Not required for full credit — just a sign of growth toward the next proficiency band.
• 3 — Target (Full Credit)
  • A 3 means the student met the expectations for the task.
  • In my gradebook, 3 = full credit.
• 2 — Developing
  • A 2 means the student attempted the task but needs more clarity or detail.
  • This earns partial credit, and I encourage a retry.
• 1 — Beginning / Needs Support
  • A 1 signals the student needs intervention.
  • Maybe the response was too short, unclear, or the student froze.

🔁 When Students Keep Getting the Same Score

If a student makes three or more attempts and keeps getting the same score, that’s when I step in.

I listen myself and decide:

• if the score is fair
• if the student needs support
• if I should override the score

Snorkl supports learning, but I remain the final evaluator.

📚 How This Fits ACTFL Proficiency‑Based Teaching

Everything I do with Snorkl is grounded in ACTFL proficiency‑based instruction. At the Novice level, ACTFL emphasizes comprehensibility, formulaic language, and successful task completion, not grammatical perfection (ACTFL, 2012).

Snorkl gives students the frequent, low‑stakes output ACTFL says they need to grow.

The 1–4 scale mirrors ACTFL performance descriptors:

• 3 = solid Novice High
• 2 = developing Novice Mid
• 1 = needs targeted support

This keeps the focus on communication, not correctness, and gives students a clear path toward the next proficiency band.

Why This Has Been Transformative

1. Students actually speak every day

Even my quietest students, the ones who freeze in front of peers, now speak confidently to Snorkl. It’s private, low‑pressure, and forgiving.

2. Feedback is immediate and meaningful

Students don’t wait days for me to grade recordings. They get guidance in the moment, when it matters.

3. The rubric reflects my values, not a punitive system

Because I trained Snorkl with my own rubric language, it evaluates students the way I believe language learning should work.

4. It makes the job sustainable

Snorkl handles the repetitive listening and initial feedback so I can focus on:

• modeling language
• supporting struggling learners
• building community
• circulating and conferring
• actually teaching

It gives me back the time that the NYCPS schedule takes away.

⚠️ A Necessary Caveat: AI Is Powerful, But Not Perfect

As much as I love what Snorkl has done for my classroom, I’m also realistic:

It’s still AI. And AI can make mistakes. Sometimes a strong Novice High student suddenly gets a “1.” Sometimes multiple students get unexpectedly low scores. Sometimes the AI misinterprets an accent or background noise. That’s when I step in. I listen, I override, and I support.

Snorkl can listen to 30 students at once — but I understand their personalities, growth, and linguistic backgrounds.

AI supports daily practice. I ensure fairness, accuracy, and humanity.

🌆 Why This Matters for NYC Students

NYC classrooms are multilingual, vibrant, and full of students who deserve daily opportunities to use the target language. But the system we work in, short periods, large classes, constant transitions, makes sustained speaking practice incredibly difficult to manage.

Snorkl bridges that gap.

It gives every student a daily digital speaking coach, and it gives me the bandwidth to be the teacher I want to be, not the teacher drowning in grading.

🎤 Final Reflection

As a first‑year teacher, I knew I couldn’t personally listen to and grade every speaking sample every day. Not with three preps. Not with 45‑minute periods. Not with the pace and demands of NYCPS.

But with Snorkl, I don’t have to choose between:

• giving students the speaking practice they deserve OR
• preserving my time, energy, and mental health

For the first time, I can confidently say: Every student in my classroom speaks the target language every single day — and they get feedback that actually helps them grow.

And that has changed everything.

References

American Council on the Teaching of Foreign Languages. (2012). ACTFL proficiency guidelines 2012. ACTFL.

About the Author

Keith is a high school World Language teacher in New York City teaching French and Mandarin Chinese. His approach to teaching is shaped by his own experiences as an immigrant language learner in the US and later through immersion in France with the TAPIF program and studies at Université Paris Nanterre. He completed his teacher education at Boston University Wheelock College of Education and is currently pursuing work in learning design and technology at Penn State World Campus.